Investigation on the quantitative evaluation method of coal combustion situation in O₂-CO₂-N₂ atmospheres based on dynamic artificial neural network

Due to the semi-enclosed space of underground coal seams and the implementation of nitrogen injection measures, most coalfield fires occur in multi-atmosphere environments. At present, there is little research on the combustion situations and kinetics of coal under multi-atmosphere conditions. In this paper, simultaneous thermal analysis experiments were conducted to investigate the combustion behaviors of bituminous coal and anthracite in O-2-CO2-N-2 atmospheres with different volume fractions. A new index for quantitative characterization of coal combustion situation in multi-gas mixing atmospheres was proposed based on the combustion situation intensity index and heat release rate. Furthermore, a method to predict the coal combustion situation using artificial neural network (ANN) was proposed. The results showed that the combustion rate decreased with the increase in CO2 volume fraction and the decrease of N-2 volume fraction under multi-gas mixing atmospheres. The volume fraction changes in inert gases had little effect on the ignition temperature or burnout temperature and had a greater effect on the combustion reactivity of bituminous coal. In a 15% O-2-5% CO2-80% N-2 atmosphere, the combustion could be inhibited for bituminous coal, while being promoted by anthracite. The combustion situation of bituminous coal in a 15% O-2-25% CO2-60% N-2 atmosphere was greater than that in air. Based on the proposed coal combustion situation index and the selected optimal ANN model, the coal combustion situation in O-2-CO2-N-2 atmospheres could be well predicted. This paper provides a reliable theoretical basis for judging the coal combustion spread in the coalfield fire area.